Saturday, March 29, 2008


Arduino chosen for next stage of RepRap development

We have chosen the open-source Arduino microcontroller for the next stage of RepRap development. Zach has been working on this for a while, and it gives a number of advantages:
In addition, we have stepper-driver boards for it that allow RepRap to use much less expensive and more easy to obtain stepper motors than the current ones.

We shall do a release of the entire RepRap system (hardware, firmware, and software) shortly that will be a stable platform from which people can take the current PIC microcontrollers forward if they wish.

We will then do a subsequent release with identical host software, but with the Arduino electronics and firmware. This will form the basis of our next set of developments.

Labels: , ,

Thursday, March 27, 2008


Buffering USB comms for Tommelise 2.0

I think that I've finally cracked the problem I was having with buffering USB data transfer between the PC and Tommelise 2.0's PIC 18F4550 microcontroller.

Recapping briefly, while a 3D printer needs to operate in real-time the PC that nominally manages it doesn't as a practical manner. With Tommelise 1.0, the data flow between PC and printer was quite sparse. The PC would tell the printer to start extruding at a particular xyz coordinate and to stop extruding at another. The printer would then convert that information into a huge number of instructions driving gearmotors and checking their progress with shaft encoders.

That control approach worked fairly well as long as what I was printing consisted of a set of long, straight extrusion paths. Once I wanted to print a large number of very short extrusion segments I started running into all sorts of problems.

(Read the whole story)

Friday, March 21, 2008



I have had some success in reducing HDPE warping by using a sparse fill pattern and then injecting polyurethane.

This object is only warped by about 0.24mm across its length, a bit more at the corners because they warp in both directions.

Full details here:

Wednesday, March 19, 2008


New IR thermometer arrives

The cheap IR thermometer that I bought arrived. It is a Raytek MT6. I got mine from Arizona Tools for $67.99. Testing it, it does pretty much everything my old Triplett ProTemp did.

About the only thing to keep in mind about it is it's accuracy. Its sensor is a thermopile with a detection range of 6.5-18 microns. That's very good in way and not so good in another. Rather than having a compensator circuit with a reference temperature pot it simply assumes that the ambient temperature is 23 degrees C.

(Read the whole story)

Saturday, March 15, 2008


Buffered USB running

Wow! When I set about to shift USB comms on the 18F4550 prototype board from direct input to buffered I had no idea what I was getting into.

(Read the whole story)


Milling with Balls and a Kitchenaid

OK here is the completed ball mill, all the separate parts put together and milling.

In the container on the mill are 50 10mm steel ball bearings and the lumps of anhydrous magnesium sulphate that need to be ground to a fine-ish powder.

The elastic bands round the container are to reduce slippage and help the drive roller with traction.

I will complete the milling during the week when the noise won't disrupt as many people, it being the weekend and all.

Construction details can be found here

Monday, March 10, 2008


A note about the McWire design

I know that many of you are quite taken with the McWire cartesian positioning system design and several of you are making one. I know I quite like it and am stealing several sweet design tricks off of it for use in Tommelise 2.0.

I've spotted a problem with the design that I discovered, however, which I will pass along to you.

One of the really cool things about the McWire design is the way he handles the connection between the steppers and the threaded drive rods. He has a fixed stepper mount and a flexible connection between the stepper and the drive rods. He also makes no attempt to secure the other end of the drive rod.

What that means is that his design is remarkably insensitive to misalignments. It self-corrects. That's utterly brilliant.

It is that very portion of the design that gives me trouble, not that I disagree with it. What the design implies is that any thrust generated by friction with the guide rails is going to go straight into the stepper motors. That's not really a problem with the beefy NEMA's that he uses. They can handle a substantial amount of axial thrust and have little, if any play in the drive axis.

(Read the whole story)


Ball Mill Frame Complete

Just for Forest, heres a picture of my repaired drill press. Note the 22mm Hole Saw.

The hole saw is what I used to cut the bearing supports into the wood end blocks and is a tight fit on the outer race of the Skate Bearings that support the rollers.

Note the line marked on the end of the hole saw to make the depth of the hole the same for all 4 holes.

N.B. Use a Drift and Vice or Press (If you have one) to insert the bearings. A quick and inexpensive drift is a socket from a socket set whose outer diameter is the same as the bearings outer race. This avoids putting undue pressure onto the bearings inner race and possibly popping the middle out of the bearing.

The Ball Mill Frame complete just awaiting gluing and screwing.

The rollers are spaced to be able to take a Milling Container with a diameter of 110mm. This is a very common and inexpensive size for plastic soil pipe and fittings and is what I will be using.

Note the end of the drive roller sticking through the end block.

Our food mixer will drive the Mill via a flexible coupling as it is what I have to hand and will borrow it when needed.

Sunday, March 09, 2008


Digging into the 18F4550

I appear to have got this USB-enabled 18F4550 chip to to what I want it to do. It has a rather complicated pre and post scaler scheme for the clock crystal which enables you to run it at its maximum speed of 48 MHz regardless of what kind of clock crystal you use. That's VERY handy.

You can also run the USB module and the CPU at different speeds, which lets you do high speed USB comms with a low speed CPU. I guess that's handy, too, though how escapes me. I have a natural tendency to run electronics and most machinery as fast as I can make it go. That's just a personality trait of mine, I guess.

Anyhow, I was able to get the servicing of the USB comms done with interrupts which left the vast majority of CPU cycles free to run the stepper. Once I did that I was able to get some respectable stepper speeds out of it.

(Read the whole story)

Friday, March 07, 2008


Tommelise 2.0 goes cross-platform

One of the big complaints from the core Reprap folks about Tommelise is that it is a Wintel phenomena. Some things I'm doing in my day job may well allow me to deal with that criticism.

(Read the whole story)

Monday, March 03, 2008


Ball Mill Rollers Complete

My Ball Mill rollers are now complete and I need to move onto making a frame to mount them in.

Here's the end detail on the Idler and Drive Rollers showing the bearings slipped on to the ends.

Here's the Idler Roller, Drive Roller and Flexible Coupling complete.

I have put most of the details in my blog

Mainly to avoid hogging the great work that everyone else here is doing. It's much more interesting than a Ball Mill.


More Crazy Ideas

Like some others, I'd thought about a feed mechanism similar to Vik's but never pursued it. In search for a mechanism that would also allow backing out the filament (as a ratcheting mechanism like Vik's does not), I thought about how other cylindrical things move... namely snakes. Here's a hastily drawn diagram of a snake-inspired friction-based feed mechanism.

It seems pretty obvious that the closer you could come to bending the filament through full 180 degree arcs, the more force you could apply. There's almost certainly some great math out there for averaging force vectors and friction coefficients... Anyone?


Sunday, March 02, 2008


Curiouser and curiouser

For those of you who haven't been following my little oddyssey recently, some time ago I put aside my little shaft-encoded gearmotors and decided to give stepper motors a go. Rather than use the 1.8 and 0.9 degree step angle NEMA 23 behemoths that Darwin uses, however, I decided to see if I could make do with much smaller and cheaper 15 degree tin can (permanent magnet) steppers. Smaller motors meant less amperage requirements and cheaper drive electronics.

I wanted to continue to use a threaded rod drive system, which meant that I didn't need the resolution that the small step angle steppers could give. Fifteen degree steps with a 1 mm pitch threaded rod got me a 0.042 mm resolution, more than what I needed.

Bigger steps meant that I had to use fewer steps to get places which meant that I could run the stepper motor at lower stepping rates which meant that I could get more effective torque out of a given motor.

I was able to test my first tin can stepper on my old Tommelise 1.0 controller board. It seemed to be giving me everything I had hoped for; adequate torque, great speed, good resolution and cheap, simple electronics electronics.

About that time I got my first USB-friendly PIC chips, the 18F4550 and was impressed with what it could do as well. Thus encouraged, I put away my old Tommelise 1.0 board and went at doing all my development work with the new, USB-friendly PIC board.

That was when things started to get sticky.

(Read the whole story)

This page is powered by Blogger. Isn't yours?

Subscribe to
Posts [Atom]